Household Waste-Sorting Dataset from Hai Phong, Vietnam
Description
This dataset supports the development and validation of the Environmental Culture Index (ECI) presented in the study “Developing an Environmental Culture Index for Household Waste Sorting”. It contains 445 complete, de-identified responses collected through a household waste-sorting survey in Hai Phong, Vietnam, from July to August 2025. The analytical dataset contains 74 positively oriented five-point Likert-scale items mapped to 15 fixed first-order constructs across Knowledge (K), Action (A), Utility (U), and Contribution (C). The synchronized repository includes Supplementary Files S1-S7, the Vietnamese questionnaire and corrected US-English analytical translation, supplementary results and figures, the complete three-phase code-and-results archive, and integrity records. The locked workflow evaluates 22,785 theory-admissible KAUC specifications and identifies KAUC_K00011_A011_U1000_C011 as the final reference specification (35 items, seven constructs), with tau = 2.9 and gamma = 2.0. Direct personal identifiers and timestamps have been removed. The ECI is a domain-specific diagnostic for household waste sorting, not a national-culture score or causal estimate of recycling performance.
Files
Steps to reproduce
The data were collected in Hai Phong City, Vietnam, from July 10 to August 18, 2025, under Vietnam National University project QG.24.92, using face-to-face interviews and online administration through Google Forms. The Vietnamese questionnaire contained 74 five-point Likert-scale items covering the 15 fixed KAUC constructs. Of 487 submitted records, 11 were excluded because preceding project surveys were incomplete; 29 respondents reporting no household waste-sorting activity followed the routing rule and skipped the subsequent Autonomy, Social Influence, Creativity, and Community Leadership blocks; and two incomplete pilot records were removed. The final analytical dataset contains 445 complete, de-identified responses. Names, contact details, addresses, timestamps, raw Google Forms identifiers, and other direct identifiers were removed before publication. To reproduce the analyses, download ECI_Supplementary_Package.zip and begin with README.md and Supplementary File S3 (S3_ECI_COMPLETE_3_PHASE_CODE_AND_RESULTS_ARCHIVE.zip), the authoritative numerical-replication archive. Follow its master guide and run the phase-specific Google Colab notebooks in sequence. Phase I uses the shared 445 × 74 dataset and 74-item/15-construct metadata registry and passes only strict PASS scenarios to Phase II-A. Phase II-A (archive Phase II1) performs PLS-SEM validation and passes only PASS scenarios to Phase II-B. Phase II-B (archive Phase II2) constructs and calibrates the ECI across 441 tau-gamma combinations, after which its required outputs enter Phase III validation and robustness analysis. The internal II1/II2 labels are retained to preserve existing manifests and checksums. The locked chain reduces 22,785 Phase I candidates to 1,642 strict PASS scenarios, then to 523 Phase II-A PASS specifications. All 523 pass Phase II-B, with 27 entering the priority pool; Phase III-A retains 522 of 523, and Phase III-B evaluates 37 deep-validation targets, yielding 25 PASS, 5 REVIEW, and 7 FAIL specifications. Five finalists remain after the prespecified interpretability and robustness gates, from which the final reference specification is selected. The workflow reproduces PLS-SEM diagnostics, respondent-level KAUC and ECI scores, calibration, bottleneck, sensitivity, cross-validation, robustness, and final-reference outputs. Results can be checked against the supplied reference files, manifests, and SHA-256 checksums; the corrected English questionnaire and metadata registry provide the exact item-to-construct-to-layer mapping.
Institutions
- Vietnam National University, HanoiAn Giang Province, Hanoi